ready to help
- Big Data in Numbers
- Data-driven prototype with gamification factor in banking
- Often there is a lack of a uniform data view
- How to increase competitiveness in Banking
- Ecosystems along the healthcare customer journey
- How can Big Data improve the customer experience in a healthcare ecosystem
- Outsourcing models and finding reliable partners
- Various forms for “Outside Resource Using”
- A case study example: A multinational automotive company based in Spain
- Smart farming, explained
A virtual copy
Imagine a virtual copy of a busy building: the physics underlying the structure are comprehensively mimicked by the computer program, and sensors spread throughout the building’s control system feed the simulation with real-time data about temperature, material degradation, water pressure, ventilation status, CO2 levels inside the rooms, and so on. A failure occurs, and the maintenance team receives a phone call. From a remote location, the staff can immediately start an in-depth investigation of the digital avatar, detecting the cause of the problem without having to actually check the building. This is one of the futuristic scenarios opened up by advanced digital-twin technology.
50 percent of large manufacturers will
have at least one digital twin initiative
launched by 2020, and the number of
organisations using digital twins will
triple by 2022, according to a Gartner
Digital twins have been made possible in recent years thanks to advances in sensor technology, the IoT, artificial intelligence and data science. Today, more and more companies – from the manufacturing sector to the engineering industry – are starting to use such technology to optimise products, refine processes and predict possible failures. Another example is a digital twin of an aircraft turbine – the avatar collects, analyses and monitors real-time data from the turbine’s sensors, and simulates any potential problem that might occur before it happens in reality, saving both costs and the time needed for maintenance.
It is possible to have a digital twin of a single object or of a complex system. Avatars can emulate the storing process of a warehouse, a service offer of a company, a workflow in a factory, an entire car or an airplane: you can deploy a digital twin of something as small as a lightbulb or as complex as a whole city. Extremely accurate digital twins – also called predictive twins – that embed historical data from other similar devices might also simulate how a device will perform over time, predicting future performance under different conditions as well as possible failures. A digital twin can also be created for an object that doesn’t yet exist. Take the example of a newly designed device: before one physically starts building anything, a digital twin can be used to run simulations and fine-tune the technology behind it. However, building a digital twin is not such an easy job.
How to build a digital twin
Digital twins are complex computer programs. Inputs for the program are data from a real object or system; outputs will be simulations and predictions on how the data will react according to different parameter changes. The digital twin is anchored to its real twin through sensors transmitting real-time data, so that even the smallest change occurring to the physical object will instantly affect the digital copy. To build a digital twin, you need a team of specialists, as currently there is no standardised platform for this task. Sensor experts map the real object, and applied mathematicians as well as data scientists – with strong skills in machine learning, artificial intelligence and predictive analytics – shape the simulation.
The team analyses the status of multiple components of the specific asset and the physics that underlie it, then develops a mathematical model with which to emulate the object, taking real-time data from sensors. In the best-case scenario, any information that could be obtained from a physical asset can also be obtained from its digital twin. Still, a digital twin can be as complex or as simple as needed: you can always decide how many of the thousands of physical variables to keep in the digital counterpart, thus modelling the digital twin according to the intended use.
Three steps to creating your digital twin
Beginning to create a digital twin can appear daunting, but can be broken down into three stages:
There are two main elements to the design of a digital twin: First, you need to select the enabling technology you need to integrate the physical asset within its digital twin to enable the real-time flow of data from the IoT devices and integration with operational and transactional information from other enterprise systems. You need to be clear about the type of device you require, the modeling software needed to create the 3D representation of the asset and who is going to have access to the information within the Digital Twin or gain control of the physical asset through it. Secure IoT device management is crucial for overcoming the risks associated with identifying the devices on your network. It provides the capabilities to authenticate, provision, configure, monitor and manage each device. An identity-driven IoT platform allows you to do this quickly and securely at scale. This leads to the second element in design. You must understand the type of information required across the life cycle of the asset, where that information is stored and how it can be accessed and used. It’s important that information is structured in a reusable way that can be quickly and effectively exchanged between systems. An identity-driven IoT platform can manage the identity of every element involved in the digital twin and provide messaging services to automate the secure communications between these people, systems and things.
You must decide the function of your digital twin. Is it simply for monitoring the asset? Do you want the twin to control and alter the asset? Do you want to make data from the asset available for advanced analytics to assist with predictive maintenance? Or, do you want to use the data and models within the twin to perform simulations to help with operational performance and product development? The answer to these questions will determine the types of devices you attach to the asset and whether you use more sophisticated devices that allow information processing to move to the edge. It will also determine your integration and data preparation, and will identify management requirements. The more sophisticated the application for the digital twin, the more comprehensive these capabilities. For example, most twins will look to exploit analytics to improve operational performance and decision-making. Controlling how data is ingested, stored, prepared and presented is essential to enable you to apply advanced analytics. To achieve high-quality results, you have to guarantee the quality of data coming from your IoT devices. Each IoT device, including its rights to transfer and accept data, is verified. Taking an identity-by-design approach builds these capabilities into your digital twin from the outset.
Most digital twin implementations start small, such as monitoring the performance of a single part within an asset, but expand over time. This happens in two ways. First, organization brings a number of smaller digital twins together to give a complete picture of an entire machine, asset or business process. Second, organizations add more sophisticated capabilities – such as simulations – into an existing digital twin. In either case, you don’t want to rip and replace but to layer up the functionality within the digital twin to meet these evolving requirements. You need to be able to securely add functionality to scale while maintaining performance to meet the extra data that needs to be gathered and managed.Download PDF